Understanding flexibility for multifunctional flood defences: a conceptual framework

The Netherlands is a flood-prone lowland country located in the delta of four large rivers along the North Sea coast. At present, eight million people live in the embanked areas, where roughly 65% of the country's gross national product is generated (Kabat et al. 2009). The threat of flooding is, however, subject to change. On the one hand, natural environmental changes (e.g. sea level rise and land subsidence) pose changes in the frequency and severity of extreme events. On the other hand, continuous land use alteration and economic development influence the exposure of impacted people and valuable assets (Djordjević et al. 2011). In such a dynamic situation, preserving the safety and prosperity of the embanked areas requires persistent investment in maintenance and reinforcement of flood defences (Stive et al. 2011).

Reinforcement of flood defences often requires space, which is scarce in the densely populated regions. The competing needs of housing, commerce, transportation, and agriculture have to fit in a relatively small surface area. Simultaneously, the safety of the living environment and the quality of the landscape has to be maintained (Ligtvoet et al. 2009). One strategy to address this issue can be by co-locating several activities in the available space (Woltjer & Niels 2007). This can be achieved by integrating urban functions into the flood defences, and is referred to as multifunctional flood defences (MFFDs) (Van Loon-Steensma & Vellinga 2014). MFFD development can increase the synergy between reinforcement and urban development by maintaining sufficient safety and enhancing the quality of the living environments along with having a lower total land requirement (Stalenberg 2010).

The resulting artefacts are long-lived, capital intensive and irreversible investment interventions. The performance requirements for these artefacts can vary considerably due to socio-economic developments, technological evolutions and natural environmental changes. Since choices made today will influence those of tomorrow, inflexibility can lead to inadequate system performance with unnecessary capital and operational costs or the need for expensive system upgrades to meet the future requirements (Ajah 2009). Conversely, flexibility is a desirable feature that can enhance the system capabilities and functionality (Schulz et al. 2000), and lessen the effects of erroneous decisions throughout the entire life-cycle (Gersonius et al. 2013).

The use of flexibility in water management for coping with extreme events is not a new topic of discussion (Gunderson 1999; Olsson 2006; Colombo & Byer 2012). It has been long realized that the deterministic and probabilistic forecasts of extreme events based on historical records do not provide sufficiently valid information for decision making (Milly et al. 2007). Furthermore, uncertainties about climate change impacts and socio-economic developments do not allow for precise quantification of potential damages caused by weather extremes (Tol 2005). Lack of definite information motivates the shift from a ‘predict and control regime’ to a more flexible management approach based on learning over time (Pahl-Wostl 2007). In this regard, flexibility is generally perceived as positive, valuable, and advantageous to have, though there are limited publications that focus on identifying and evaluating it in the field of water management (Fankhauser et al. 1999; Linquiti & Vonortas 2012; DiFrancesco & Tullos 2014). Concurrently, there exists no generally accepted characterization and definition of flexibility throughout the literature.

It is the purpose of the present paper to develop a conceptual framework to clarify the concept and structure the discussion of flexibility in the context of MFFDs. This study is seen only as the first step in the process of developing a methodology for identifying and evaluating flexibility in the design and management of MFFDs. The research method follows a positivistic approach (Lin 1998; Warfield 2005) and focuses on a theoretically based identification of the commonalities that hold across different fields in order to be used as the building blocks of the proposed conceptual framework of the study. The paper is divided into three major parts as explained below.

First, the concept and usage of flexibility is reviewed as presented in a variety of literature. Over three hundred worldwide academic publications have been explored initially. The selection of the literature began with Seebacher & Winkler (2013), who have conducted a citation analysis of flexibility literature in manufacturing, and was extended by using a Google Scholar search on a wide range of flexibility-related keywords.

Second, the conceptual framework of the study is proposed in the form of four self-consistent and step-wise questions. The questions are adapted from Upton (1994) and address the commonalities found in the literature. A selection of thirty-five publications is then synthesized to distil the eight characteristic features of flexibility, in association with the proposed questions. Subsequently, a working definition of flexibility is proposed as well.

And third, an illustrative case study follows to demonstrate the application and potential of the framework in discussing, identifying, and evaluating flexibility for the development of an MFFD. The paper ends with some challenges and recommendations for future research.

A necessary step for better understanding of flexibility in the context of MFFDs is to remove the ambiguity and vagueness of the concept. Indeed, much can be done by investigating the insights presented in the literature. Flexibility is a topic of interest and discussion in various fields, such as product and organizational design (Sanchez & Mahoney 1996; Kandemir & Acur 2012; Singh et al. 2013; Kok & Ligthart 2014), information technology (Byrd 2000; Highsmith 2013; Nanath & Pillai 2014), business development (Bryson et al. 1993; Regev et al. 2006; Arnold & Artz 2015), infrastructure development (Zhao & Tseng 2003; De Haan et al. 2011; Gil & Biesek 2014), adaptation to climate change (Few et al. 2004; Adger et al. 2005; Heller & Zavaleta 2009), and complex systems (Ethiraj & Levinthal 2004; Alkemade et al. 2009; Moses 2010).

The literature review of this study is, however, limited to three areas of investigation. The field of flood management is chosen because it is the primary field of interest. Moreover, during the initial investigation of the worldwide literature on the web, it was observed that the generic principles of flexibility as presented in the real options and manufacturing publications have provided the foundation for conceptualizing and operationalizing flexibility in various fields such as aerospace systems (Galbraith 1990; Saleh et al. 2003; Nilchiani 2005), information technology (Duncan 1995; Panayi & Trigeorgis 1998; Dorsch 2015), infrastructure planning (Ajah 2009; De Neufville & Scholtes 2011), port development (Mansouri et al. 2010; Taneja et al. 2014), water supply and waste water systems (Zhang & Babovic 2012; Spiller et al. 2015), emergency management (Ward et al. 2015) and urban planning (Geltner & De Neufville 2012; Macintosh 2013). Hence, it was assumed that the insights offered by real options and manufacturing publications may also provide sufficient coverage for understanding flexibility for MFFDs.

This section aims to present the varied conceptual usage of flexibility as discussed in the literature. The multi-disciplinary literature review of the concept of flexibility is not meant to be exhaustive. Rather, the chosen publications are considered to represent significant contributions to a wide body of knowledge about flexibility in the areas under consideration. The end-goal is to derive the major themes and limitations within each context, and the commonalities and inconsistencies across the flood management, real options, and manufacturing literature.

The changing threat of flooding is a major concern for the development of flood protection measures (Downton et al. 2005). On the one hand, climate change is expected to alter the frequency and severity of extreme events (IPCC 2007), which is of particular relevance for flooding (Djordjević et al. 2011). On the other hand, demographic and economic developments in the floodplains are among the non-climatic factors that influence the consequences of floods (Nicholls 2004). Climatic and non-climatic changes are interlinked (Zevenbergen et al. 2008), and non-stationary in their nature (Gersonius et al. 2013). Consequently, not only the magnitude and speed of changes in flood hazard, but also the extent of the consequences of flooding are deeply uncertain and unpredictable (Rahman et al. 2008).

Provision of sufficient safety in such a dynamic environment is a dilemma. On the one hand, flood defences are ‘quasi-irreversible’ (Fankhauser et al. 1999) investment interventions that cannot be easily upgraded. On the other hand, the performance requirements are expected to change in accordance with the environmental changes. In anticipation of change, consideration of flexibility in the design and management of flood protection measures is increasingly recommended in the literature (Nicholls & Branson 1998; Adger et al. 2005; Gersonius et al. 2011; Woodward et al. 2011). Achieving flexibility proactively is expected to be cheaper than reacting to change after the occurrence (Stern 2006; Tol et al. 2008). Moreover, enhancing flexibility can ensure proper performance of the flood defences under a range of plausible future conditions (Smit & Pilifosova 2003).

A commonly cited approach to achieving flexibility for a typical flood defence is by allowing for mid-term adjustments and modifications of the structure according to the new insights gained over time (Klein et al. 1999; Morris et al. 2007; Tol et al. 2008; Van Buuren et al. 2013). The investment ‘options’ are left open for future adaptation (Haasnoot et al. 2012), and are postponed until the time that the costs of further delay are greater than the benefits (Felgenhauer & Webster 2013). Several studies have shown the added value of embedding investment timing flexibility through application of real options techniques (Dobes 2009; Scandizzo 2011; Woodward et al. 2011; Linquiti & Vonortas 2012; Jeuland & Whittington 2014).

A growing body of literature discusses the value of incorporating real options in the long-term development planning for capital intensive projects under market uncertainty. The term ‘real option’ was first used by Myers (1977) referring to the ‘right, but not an obligation’ to modify the system under consideration to adapt to its changing environment (Cardin & De Neufville 2008). Having the right to revise the decisions on a predetermined cost and at any time adds value to the option and makes it distinctive from an ‘alternative’ or ‘choice’ (De Neufville 2002). The option pricing techniques are used to compare the costs of delaying decisions and acquiring flexibility with the benefits of waiting (Scandizzo 2011).

Various types of real options – such as the option to defer, stage, or expand the investment – have been proposed in the literature (Amram & Kulatilaka 1999; Trigeorgis 2005). The options are aimed to provide flexibility for hedging against the negative impacts of uncertainties and take advantage of unexpected upside opportunities created in the future (De Neufville & Scholtes 2011). Each option creates an opportunity for a managerial decision or action that can be taken in response to the changes in the market. Management plans are considered flexible if they can be delayed and updated periodically and proactively (Luehrman 1998).

The focus of the options thinking approach is on measuring the financial value of embedding managerial flexibility. The method is limited to market uncertainty as the only source of uncertainty. However, the proposed options are comprehensive and can be used in various investment projects. The different aspects of flexibility are only partially discussed.

There exist a large number of publications discussing flexibility in manufacturing (Slack 1983; Gupta & Goyal 1989; Mandelbaum & Buzacott 1990; Koste & Malhotra 1999; Duclos et al. 2003). The need for flexibility in manufacturing is related to the complexities created by technological advancements, rapidly changing business environment, constant pressure to upgrade the products, and satisfying customers' preferences (King & Sivaloganathan 1999; Shi & Daniels 2003; Rajan et al. 2005). In response, flexibility is aimed at demand management, shortening the lead times (Slack 1983), quickness of response (Fisher et al. 1994), and responsiveness (Holweg 2005). Flexibility is generally understood as a system capability to be reconfigured in order to hedge against uncertainty (Van Mieghem 1998; Goyal & Netessine 2007), and to maintain profitability and competitiveness (Zelenović 1982) in a cost efficient way (Duclos et al. 2003).

Since different uncertainties exist, various types of flexibility have been identified and discussed within the published literature. They vary from adjustments in the system components (e.g. labour, material, machine), organization (e.g. procedure, processes, volume), products (e.g. production, market), and distributions (e.g. responsiveness, network) as identified by several scholars such as Browne et al. (1984), Gerwin (1993), and Duclos et al. (2003). However, the types of flexibility have been interpreted differently by different scholars (Seebacher & Winkler 2013). The differences are influenced by the context of the system (De Toni & Tonchia 1998), multiplicity of variables (Kersten et al. 2011), and multidimensional nature of flexibility (Sethi & Sethi 1990).

Subsequently, different logics have been used for classifying the growing number of flexibility types into different dimensions. For example, Bernardes & Hanna (2009) and Eppink (1978) use the nature of uncertainty and predictability to classify flexibility into reactive (passive) or proactive (active) responses; Zelenović (1982) was one of the first to characterize flexibility based on the time frame dimension of change into operational, tactical, and strategic; Upton (1994) used the dimensions of ‘internal’ and ‘external’ flexibility based on the intention of flexibility to accommodate competitiveness via internal capabilities of a manufacturing unit (‘what we can do’) or adjusting to the external advantages derived from it (‘what the customer sees’). Although the manufacturing literature provides a broad view of the flexibility concept, there is no general agreement on the definition and characterization of flexibility within the field. Moreover, the classification schemes proposed are fairly unstructured, fragmented, and complex, and need to be adjusted to the context under consideration (Upton 1994; Winkler & Seebacher 2012).

Concluding from the literature review, it can be seen that throughout the literature, the need for flexibility is related to changing circumstances, though the nature of change and degree of uncertainty about the change vary by context. Second, flexibility is seen as an advantageous and desirable attribute capable of handling uncertainty and change. However, the preferred goal and the capabilities of flexibility in achieving them are specified differently based on the context of the problems faced in each field of investigation. Third, flexibility often entails a kind of response to uncertainty and change, although what characterizes the response is described by disparate dimensions depending on the nature of change and context of the system under consideration. Fourth, each field of investigation proposes some ways of achieving flexibility. However, they differ widely in relation to the nature of change and uncertainties and their impacts. Overall, the observed commonalities demonstrate a certain level of consistency in conceptualizing flexibility across the literature. However, the field-specific and context-based characterizations of these common aspects create the sources of inconsistency and ambiguity across, and within the disciplines.

The literature review has demonstrated that while the conceptualization and operationalization of flexibility is more advanced in the context of real options and manufacturing, in the field of flood management the concept needs to be enriched. To do so, this section proposes a conceptual framework, as defined by Maxwell (2005), to be used to clarify the concept and structure the discussion of flexibility in the context of MFFDs. The research adopts the stance of positivistic approach (Lin 1998; Warfield 2005), in which the framework is built based on the observed commonalities in the literature. Such an approach has been similarly applied by many scholars such as Taljaard et al. (2011), in the context of integrated coastal management, and Nilchiani (2005), in the context of aerospace engineering. Both studies have used the commonalities in the literature for developing their proposed frameworks.

In the first step, the observed commonalities in the use of flexibility across the literature are presented in the form of four step-wise and self-consistent questions as presented below. The questions are adapted from Upton (1994), though the intended meaning of the dimensions of flexibility in Q3 is not the same as stated by him. Rather, this research uses the word dimension as applied by Evans (1991) and Golden & Powell (1999), which supports the intention of this research. The four questions are as follows. Q1. Why is flexibility needed? This question establishes the motivation for consideration of flexibility. Q2. What is it that flexibility is required for? This question seeks to describe the competences of the flexibility concept. Q3. What are the dimensions of flexibility? This question indicates the extent to which flexibility can be achieved. Q4. What needs to change or be adapted? This question discusses the potential ways of achieving flexibility.

Each question of the framework addresses one of the common aspects of flexibility derived from the literature. Since the common aspects are consistent across the fields, they are assumed to be instrumental for structuring the discussion of flexibility for MFFD development as well.

However, the four questions help to draw the spectrum of the areas that needs to be considered in discussing flexibility. The observed sources of ambiguity stem from inconsistency in characterizing flexibility across the literature. In order to provide greater clarity and to answer the questions without confusion in the context of MFFD, the characteristic features of flexibility associated with each question need to be clearly specified.

The review of over three-hundred flexibility-related publications demonstrated that the majority of the publications are focused on operationalizing and implementing (Boyle 2006; Ewert et al. 2009; Hallegatte 2009; D'Angelo et al. 2013, Filatova 2014), measuring (Ramasesh & Jayakumar 1991; Dixon 1992; Georgoulias et al. 2007; Moon et al. 2012), and evaluating (Ito 1987; Kulatilaka 1988; Cortazar et al. 1998; Christopher & Holweg 2011; Chod et al. 2012; Linquiti & Vonortas 2012) flexibility. This section selects and synthesizes 35 papers that focus on presenting a self-definition of flexibility, developing distinguishable characteristic features of flexibility, and/or presenting a thorough review of the preceding publications in the field.

A best-fit approach (Bond 1994) is taken to distil the characteristic features of flexibility. The attempt is to select the characteristic features that are meaningful in both real options and manufacturing contexts. Correspondingly, the case-specific characteristic features are omitted. Hence, the chosen characteristic features are assumed to be context-independent. This assumption follows the analysis strategy outlined by Bond (1994) and Shackelford et al. (2005).

Table 1 summarizes the four questions of the framework, their associated characteristic features, and the spans of the terms used for searching the references. There was an attempt to search each reference for explicit use of the stated terms, though in some cases, in particular relevant to the second and fourth questions (Q2 and Q4), the intent has been inferred based on the implicit evidence presented in the authors' work.

The characteristic features associated with each question and their spans are detailed in the following paragraphs. The distributions of the eight characteristic features as observed in the real options and manufacturing publications are depicted in Tables 2 and 3, respectively, later in this paper. It can be seen in these tables that many authors have associated the distilled characteristic features in their work implicitly or explicitly. However, the characteristic features altogether have not been comprehensively addressed before.

The common ground on which all the authors agree is the inevitability of change in the system itself and of its environment over time (Zelenović 1982; Golden & Powell 1999; Stevenson & Spring 2007). The degree of change can vary from being planned and predictable to deeply uncertain and unplanned. Both the change and uncertainty are used to characterize the need for flexibility in the literature, and are adopted in the current paper.

Sources of change can be internal or external to the system boundary (De Toni & Tonchia 1998; Van Hoek 1999; Beach et al. 2000). In the real options and manufacturing literature, the sources of external change are associated with variations in the entities such as customers, suppliers, market, and technologies (Myers 1977; Zelenović 1982; Gerwin 1993; Duclos et al. 2003; De Neufville et al. 2004). On the other hand, machine breakdowns, variability in processing times, and quality problems exemplify the internal sources of change in manufacturing (Gupta & Goyal 1989; Buzacott & Mandelbaum 2008). Twelve authors have addressed both internal and external sources of change that have a need for flexibility.

A lack of knowledge about change specifications results in admitting uncertainty (Nilchiani 2005). All of the authors have mentioned flexibility for dealing with the uncertainty associated with a source of change. In addition to the explicit use of the term uncertainty, the other terms which have been used by the authors to explain uncertainty include ‘unforeseen’ (Golden & Powell 1999; Roberts & Stockport 2014), ‘unpredictable’ (Gupta & Goyal 1989; Angkiriwang et al. 2014), and ‘unplanned’ (Correa & Slack 1996) changes. However, the manufacturing literature considers flexibility for handling both types of predictable and unpredictable changes (Golden & Powell 1999, Beach et al. 2000). The real options literature claims that as the severity of uncertainty increases, it is more preferable and worthwhile to incorporate flexibility (Triantis 2003; Cardin & De Neufville 2008). Here in this paper, in addition to the change, uncertainty is the second characteristic feature that is adopted for discussing the need for flexibility.

Here, the primary attempt is made at establishing what will be offered by taking flexibility into account. This can be described by the two characteristic features of the goal and capabilities of flexibility. The former demonstrates the desired end result of having flexibility, whilst the second demonstrates the abilities of flexibility to achieve its goal.

Flexibility is aimed in the literature at handling both the downside and upside of uncertainty and change. The manufacturing literature represents an explicit focus on maintaining and enhancing competitiveness and profitability in a cost-effective way (Mandelbaum & Buzacott 1990; Bernardes & Hanna 2009; Jain et al. 2013), which is more prone to make a benefit of uncertainty and change. The handling of the negative consequences of uncertainty and change has also been mentioned by several authors (Eppink 1978; Sethi & Sethi 1990; Da Costa 2012). However, in the real options literature flexibility is always aimed at both capitalizing favourable future investment opportunities and hedging the risks (for example, Wang & De Neufville 2006; Cardin 2014). There are, in total, 23 authors who have addressed flexibility for coping with both downsides and upsides of uncertainty and change, implicitly or explicitly.

Flexibility is often qualified based on the capabilities offered by the consideration of flexibility (Gupta & Goyal 1989). Capabilities of flexibility can be described in terms of ‘scope’ and ‘achievability’ (Bernardes & Hanna 2009). The scope refers to the total number and range of options the system can accomplish whilst options are held in reserve to meet the future needs (Slack 1983; Mandelbaum & Buzacott 1990; Gerwin 1993; Volberda 1996). Achievability denotes the ‘ease’ of transition (Sánchez & Pérez 2005), transition ‘penalties’ (Upton 1994), or ‘mobility’ in terms of cost, time, or performance losses (Koste & Malhotra 1999) for attaining each option within the scope. The quantified capabilities of flexibility are used as indicators of cost-effectiveness, and efficiency of flexibility consideration (Slack 1983; Gupta & Goyal 1989; Gerwin 1993; Beach et al. 2000). Furthermore, there are 18 authors who have addressed both the scope and achievability.

This study chooses the characteristic feature of ‘goal’ of flexibility to not only reduce the downside losses and vulnerabilities, but also to exploit the upside opportunities created in the future. Besides that, the selected characteristic feature of ‘capabilities’ involves both the scope (the range/number of options that can be achieved) and the achievability (the transition time, cost, and performance losses) for qualifying flexibility.

Flexibility is acknowledged to be multidimensional (Sethi & Sethi 1990) and polymorphous (Evans 1991). The literature on dimensions of flexibility is rather vast and articulated (Seebacher & Winkler 2013). There is inconsistency in the use of the term dimension even within a single firm (Golden & Powell 1999). Similar to Evans (1991), and Golden & Powell (1999), the current paper uses the term dimension to indicate the extent to which flexibility can be achieved. Accordingly, the two dimensions of the ‘mode of response’ and ‘temporal’ are derived from the literature, which are found to be generic and independent of the context of the system under consideration. Hence, these two are chosen for characterizing the dimensions of flexibility for the current paper.

The mode of response indicates the standpoint of decision makers towards flexibility based on the effects of change (Golden & Powell 1999). A defensive mode of response represents a passive reaction to change after the occurrence, and aims at minimizing the negative impacts and losses (Eppink 1978; Evans 1991; Gerwin 1993). It is an event-driven approach, and it relies strongly on the combined use of control and buffers (Da Costa 2012). The offensive approach actively monitors the change process and its impacts. The goal is to prevent the negative impacts and take advantage of the created opportunities in anticipation of external changes (Gerwin 1993; Triantis 2003). The system configuration is then proactively redesigned and altered in anticipation of change (Angkiriwang et al. 2014). The manufacturing literature makes use of both response types to respond to foreseeable and unforeseeable changes (Eppink 1978). However, the real options literature encompasses various forms of proactive management of uncertainty (De Neufville et al. 2004). There are 16 authors that have addressed both reactive and proactive modes of response, which are both considered as components of the characteristic feature of ‘mode of response’ for the framework of the current paper.

The temporal dimension reflects the period of time over which change will happen (Zelenović 1982; De Toni & Tonchia 1998). The time horizon of change can vary from short term to long term, while the frequency of change can be discrete or continuous (Upton 1994). Three widely established categories of the temporal dimension of flexibility are operational, tactical, and strategic (Eppink 1978). Operational flexibility entails rapid reaction to short-term, discrete and predictable changes such as machine breakdown or a shortage of raw material. It includes a range of operations the system can handle without a major setup (Upton 1994; De Toni & Tonchia 1998; Stevenson & Spring 2007). Tactical flexibility entails occasional system alternation with some effort and commitment. For instance, a major opportunity for improvement without changing the overall system configuration exemplifies tactical flexibility (Upton 1994; De Neufville et al. 2004). Strategic flexibility denotes long-term changes in response to the continually changing environment. It involves dynamic alteration in the design, development, and operation of the system (Roberts & Stockport 2014). Although most of the non-marked authors have addressed the strategic flexibility (for example, Slack 1983; Amram & Kulatilaka 1999; Trigeorgis 2005; Angkiriwang et al. 2014), only 11 authors have mentioned all the three strategic, operational, and tactical time frames. These three time frames are all used to represent the span of the temporal characteristic feature of flexibility for MFFDs.

In addition to the temporal and mode of response dimensions, there are other dimensions of flexibility defined in the literature that are not considered for characterizing flexibility here in this paper. For example, De Toni and Tonchia (1998) propose three dimensions of ‘horizontal’, ‘vertical’, and ‘by the object of the variation’. The horizontal and vertical dimensions of flexibility refer to the phases of manufacturing and levels of hierarchy in manufacturing operations. The object of variation outlines the location of flexibility in relation to the boundaries of manufacturing. The dimension of ‘by the object of the variation’, has also been called ‘internal’ and ‘external’ flexibility by some authors such as Eppink (1978) and Upton (1994), and ‘focus’ by Golden & Powell (1999). The internal and external flexibility show that flexibility is not confined by the organizational boundaries of manufacturing. For example, trading relationships can extend the flexibility of manufacturing (Golden & Powell 1999). The dimensions discussed in this paragraph are specific to the nature of manufacturing. Therefore, they are not used to characterize flexibility for the framework of the current paper.

The last question of the framework looks for the ways of achieving flexibility or the sources of flexibility. In this paper, the two characteristic features of flexibility ‘types’ and ‘enablers’ are adopted to describe the potential ways of achieving flexibility as defined by Mikaelian et al. (2011) and Cardin (2014).

The approach of the real options literature towards determining the ways of achieving flexibility is comprehensive. Basically, the real options literature refers to a variety of managerial decisions and actions that can be taken to respond to change and uncertainty (Myers 1977; Triantis 2003). These managerial options are independent of the context of the system and can be applied for any investment intervention under uncertainty. Often, the conventional real options literature does not address what enables the managerial options. More recently, Mikaelian et al. (2011) and Cardin (2014) have characterized the ways of achieving flexibility into the flexibility ‘types’ and ‘enablers’. The flexibility types or real options ‘on’ a system (Wang & De Neufville 2006) have the same meaning as the conventional real options. Some examples of these real options are the options to expand, defer, and shrink the investment (Trigeorgis 2005). Flexibility enablers are also called ‘design flexibility’ (Saleh et al. 2003), flexibility ‘in’ the project (Wang & De Neufville 2006), and the flexibility ‘mechanism’ (Mikaelian et al. 2011). Flexibility enablers refer to the sources of alterations in the technical design of a system to make it changeable, and are determined based on the nature and effects of uncertainty and change (Cardin & De Neufville 2008). Except Wang & De Neufville (2006), Cardin & De Neufville (2008), Mikaelian et al. (2011) and Cardin (2014), the other marked authors from manufacturing (in Tables 2 and 3) have not made a distinction between the flexibility types and enablers. Rather they have mentioned various samples of flexibility types and enablers implicitly.

It should be noted that in the context of manufacturing, flexibility is often conceived of as hierarchical (Sethi & Sethi 1990; Koste & Malhotra 1999). The underlying assumption is that the flexibility of sub-components (e.g., technology, human resources, and supply networks) contributes to the overall system flexibility at a ‘higher-level’ (Volberda 1996; Koste & Malhotra 1999). Ways of achieving flexibility are then classified based on the tier to which they belong (Sánchez & Pérez 2005). For example, Koste & Malhotra (1999) map the sources of flexibility for manufacturing into four tiers of ‘individual resources’, ‘shop floor’, ‘plant’, ‘functional’, and ‘strategic business unit’. Since the flexibility hierarchy does not fit in the nature of MFFDs, this approach has not been taken into account in the current paper.

It is argued that the proposed framework, including four questions and eight characteristic features, can be instrumental in structuring the discussion of flexibility in the context of MFFDs for two reasons. First, the four questions of the framework represent the commonalities observed in the literature that are consistent across the flood management, real options, and manufacturing publications. Second, the terms applied for characterizing flexibility have been sporadically applied in discussion about adaptation to climate change (Fankhauser et al. 1999; Smith et al. 2000; Adger et al. 2005). Hence, the terms are known and can also be used in the context of MFFD development.

However, the framework is not claimed to be universal and comprehensive. It is aimed to be useful for different actors and stakeholders involved in decision making and development of MFFDs. The end goal is to improve the overall effectiveness of the project in accordance with their perspectives and interests. Thus, the following working definition of flexibility is proposed:

This section demonstrates the developed framework by applying it to the conceptual design of an MFFD. The investment interventions required for the development of an MFFD are capital intensive and irreversible. Hence, the framework is intended for planning the structure, taking into account the dynamics in the environment and changing circumstances. The objective of the process is to support the inclusion of flexibility in the design and management of MFFDs in such a way as to not only meet today's requirements, but also to accommodate future needs.

The illustrative case study of this section is anonymous. The context for the case study application is taken from an existing MFFD in the Netherlands, in which a series of residential and commercial buildings have been built on top of a sea dike. However, the information about the development process of these structures is not available to the authors. The context can serve as an example of a situation in which various types of uncertainties and changes associated with the development of an MFFD need to be addressed.

The sea dike of the case study represents a man-made, earthen structure that is in place to protect the hinterland areas from high sea water levels and wave attacks. The dike is assumed to be a section of the whole flood defence system, which protects an urban area along the coast. The profile of the dike is considered to be the same all along the section. In the Netherlands, the strength and stability of the dike sections are checked periodically to maintain the required level of safety. Presumably, the latest visual inspection of the dike has demonstrated that the dike section under study does not comply with the current safety standards and has to be reinforced. The decision on the extent of the reinforcement, however, is faced with uncertainty about various influencing factors.

In this example, the reinforcement of the section implies increasing the width and height of the dike. The extra space required has also been planned for the development of a residential area. To deal with the conflict of reinforcement and urban development, it has been suggested to construct buildings on top of the dike, with the same length as the dike. It is, however, very difficult to determine the size (the number of floors) of the high rise buildings because the demand for houses is not predictable, especially when they are integrated into a dike.

The dike and buildings are capital intensive and irreversible interventions. In anticipation of various uncertainties about the future developments, it has been realized that it is favourable and perhaps more cost-effective to incorporate flexibility in the planning of the coupled structure in this example. In order to explicitly address the desire for ‘flexibility’, it is necessary to provide a common basis for discussing flexibility without ambiguity and confusion. To do so, the developed framework is used to structure the discussions with the aim of investigating the design considerations for the dike and buildings. The framework is represented as shown in Table 4. Q1 to Q4 are the four questions of the framework. Each column of the table details the associated characteristic features of flexibility for the relevant questions. The intention is to determine characteristic features of flexibility that are as clear, unique and independent as possible. Completeness is not attempted, although each box should provide sufficient information for handling the problems faced.

Within this case study, the framework is first applied to discussing flexibility for the dike and the building developments individually. Previous publications are used to derive the necessary information for each step of the framework. Afterwards, the options for flexible design of the two structures are combined, and the impacts of the coupling on their flexibility are further explored.

Table 5 summarizes the characteristic features that have been discussed above.

With the coupling of the two structures, some challenges and complexities emerge that impact the flexibility considerations. With the developed framework, the challenges created relevant to each characteristic feature are tractable. Some of the challenges relevant to each question of the framework are exemplified here.

Regarding Q1, so far the uncertainty and changes are treated separately for the two structures. However, the coupling of the two structures may lead to interference of changes in one structure with the performance of the other. For instance, at each period within which the buildings are raised, some additional weight is added on top of the dike. This extra weight may cause dike instability. As the demand is uncertain, the degree of impact will be uncertain, but important. Therefore, demand uncertainty has to be taken into account for the design of the dike as well as the design of the buildings itself.

Regarding Q2, an interesting fact that can be inferred is that a high rise building adds some extra height to the dike. If the series of buildings on top of the dike are designed in such a way as to not allow high water levels to pass through to the hinterland, the dike can be built lower. This in turn means that constructing the buildings with water retaining walls can contribute to delaying the reinforcement investments for a longer period of time, and therefore enhances the flexibility in reinforcement planning.

Regarding Q3, the time span of change and mode of response are the factors that can impact the frequency and timing of adaptation sessions. For example, if the dike is raised then the ground floor of the buildings cannot be used any more. To comply with the need for this evacuated space, used for dike heightening, it might be necessary to add to the number of floors even if the housing demand has not changed. This implies that the timing of the decision for modifying the buildings will not only be dependent on the demand changes, but also the changes in the sea water level.

Relating to Q4, as a result of the challenges discussed above, it can be inferred that the process of determining the flexibility types and enablers has to be recursive, and iterative. It should begin with identifying flexibility for the dike and buildings, determining the various ways in which the options for flexibility can be combined, assessing the feasibility of each option, and selecting among them.

The framework of the paper was used to structure the discussion of flexibility in the conceptual design of an MFFD. It can be seen that use of the framework facilitates identification of potential flexibility attributes as well as handling the complexities created by coupling the two structures through an iterative process.

Furthermore, planning for coupling the dike and the buildings requires collaboration between the people engaged in the design and management of the dike and the buildings. To prevent mutual misunderstanding, it is of the utmost important to ensure that people with different backgrounds have the same understanding of the subject under discussion. The step-by-step application of the framework for the case study demonstrates that is possible to structure the flexibility discussion for the dike and building in the same way and without confusion.

Additionally, the characteristic features of flexibility can support the evaluation process as well as the identification process in two ways. First, the extent to which flexibility achieves its goal can be used for assessing the added value of flexibility as applied by some scholars such as Woodward et al. (2011) and Linquiti & Vonortas (2012). Second, the quantified capabilities of flexibility are often applied as indicators for measuring the degree of flexibility provided by a specific option. Those indicators can be used to compare and prioritize the developed options for flexibility (Gupta & Goyal 1989).

Overall, it can be claimed that the four questions and eight characteristic features enhance clarity about the concept of flexibility, and can serve as a first step for developers of MFFDs to formulate clear plans for identifying, evaluating and enhancing the critical flexibility required.

This paper was prompted by the observed ambiguity about the meaning and characterization of flexibility in the flood management literature. Subsequently, a framework was developed aimed at enhancing consistency and clarity in discussing, identifying and evaluating flexibility for the development planning of MFFDs. The framework consists of four questions and eight characteristic features of flexibility. The questions address the consistent commonalities found in the literature and are used to structure the discussion of flexibility. In order to clarify flexibility for MFFDs, the eight characteristic features of flexibility in association with the four questions were distilled from a selection of 35 publications from real options and manufacturing. However, many scholars have taken the identified characteristic features in their work implicitly or explicitly into account. We have shown through tables that the characteristic features altogether have not been comprehensively addressed before. The paper argues that the proposed framework, including the four questions and eight characteristic features, can be instrumental in structuring the discussion of flexibility in the context of MFFDs.

Having developed the framework, the functionality and potential of the framework were explored for an illustrative case study. The case application showed that the framework can indeed be used for structuring the discussion of, and preventing confusion and fuzziness about the meaning of flexibility. This is of particular importance for MFFD developers to have a common ground for communicating about flexibility, since they come from different disciplines. Furthermore, the framework illustrated the relationship between the flexibility and design considerations for MFFDs. Additionally, areas that need more attention in the discussion about flexibility were outlined by the framework. It was concluded that the framework can be used by developers of MFFDs as a step-by-step guideline to formulate clear plans for identifying, evaluating and enhancing the flexibility critical for MFFDs.

The study has raised several issues that should be further addressed in future research.

First, the impacts of multifunctionality on the required flexibility for the flood defence deserves further clarification. Further research should address the interactions between the dike's performance and its secondary functioning in relation to the required flexibility for MFFDs.

Second, this study is the first step towards identifying and evaluating flexibility for MFFDs. Further research is required to develop an evaluation model for assessing the cost-effectiveness of flexibility considerations in developing MFFDs. Use of social cost–benefit analysis methods can be advantageous for a systematic and cohesive assessment of a broad range of impacts caused by flexible design and development of MFFDs.

Third, the illustrative case study demonstrated the applicability of the proposed framework of the paper. There is a need to explore the instrumentality of the framework for real cases. This can be done by conducting targeted interdisciplinary workshops and interviews. Also, it is interesting to test the applicability of the framework in other non-Dutch cases worldwide.

Finally, further research is required to explore the literature in areas other than real option and manufacturing. Fields such as business development, complexity science, or urban development may potentially contribute to further revision and refinement of the framework and the distilled characteristic features.

The research described in this paper is sponsored by the Netherlands' Technology Foundation (STW).